The Future of Power Generation and Data Centers
Topics: Energy Markets & Oil's FutureEngineering and Technology
The Cloud Runs on Natural Gas
Everyone understands the carbon footprint of driving a car or getting on a plane. What gets missed, Josh said, is that the internet runs on the same hydrocarbons. Cloud storage, the social media apps on your phone, the photos sitting in a data farm — all of it draws power, and in the data center world a large share of that power comes from natural gas. It doesn’t have to be machine learning or AI to matter. Even static storage of your pictures requires power, and that demand has moved well past the footprint people associate with their vehicles.
David Abshire framed it as the light switch problem: most people flip one without thinking about where the energy comes from. With data center construction taking off in the U.S., that load is a growth path for the industry — a reliable, clean source of power for facilities that never stop drawing.
Modular Plants, Built in Blocks
The problem was speed. Data centers need power, and they need it fast, so the design work Josh described was organized around what can actually be delivered. Reciprocating engines top out around four megawatts each. Natural gas engines get larger. Combined cycle — gas turbines with heat recovery and steam generation — is what utility-scale plants run.
The modular package he worked on was built around reciprocating engines, largely because of availability. At a high level the system isn’t complicated: hook a group of machines to an electrical system and run them together to reach large-scale power, on the order of 100 megawatts and up. From there you add blocks as equipment becomes available rather than waiting on a single large order. The design was solving for that power bottleneck specifically.
From a Refinery Co-op to Gigawatt Projects
Josh’s first exposure to the industry came during college, when he co-oped at a Marathon oil refinery. What hooked him was the complexity. “I fell in love with the complexity,” he said. He also liked that the work mixed white collar and blue collar — time around people and equipment instead of behind a desk.
Today the projects run from a few megawatts to an entire gigawatt, much of it supporting hyperscale data centers. The last 12 to 18 months have brought a wave of project opportunities, he said, and electrical engineering is far more prevalent on the power generation side than it was in oil and gas.
The two ends of his experience also taught him how differently decisions get made. At a large refining facility, procedures, sign-off and checking govern everything. Safety is paramount, planning is slower, and there are plenty of models to throw at a problem. At a smaller firm, the speed of a project is on a different magnitude. Less sign-off, fewer specs and approvals, and a facility that carries less risk than a refinery. The result is a constructed facility, faster.
The Highlight Reel Problem
Attracting the next generation into field operations and skilled technical roles is getting harder, and Josh points at social media. People see the highlight reel on LinkedIn — work-from-home jobs, high salaries — and steer their careers toward what shows up on the screen. FOMO is doing the rest. “It’s just what they see on the internet,” he said. The industry’s problem is that field work doesn’t photograph well, and the pipeline of people willing to commit to it thins out a little more each year.
Eighty Percent Right Is Good Enough
Asked what he would fix with a magic wand, Josh didn’t hesitate: speed up decision making. In hyperscale work, faster to energize is the whole game, and customers will spend money to add resources and strip out bureaucracy to hit a construction timeline. Engineers, he said, like to over-analyze and walk down every path available to them.
“If you can design something that’s 80% correct, that’s gonna be good enough, probably 98% of the time,” he said. “That’s usually good enough to make a decision and move on.”
David called it analysis paralysis. Sitting on a decision for a week or two can slow a project by months — and that is a cost nobody puts in the estimate.
Terms explained in this episode
Full transcript · 1,893 words
Machine-generated transcript of “The Future of Power Generation and Data Centers”. Paragraph breaks mark a change of speaker. Click a timestamp to play from that point.
0:02 This is Under Pressure, Compressor Talk by Midwest Compressor. Strategies, systems, and stories from the compressor world. With your host, David Abshire. And now, on to the show.
0:20 Good afternoon. This is David Abshire with the Under Pressure Podcast. Today we have Josh Heff with us. Josh, thanks for joining us today.
0:29 Yeah, thanks, David.
0:33 Absolutely. So, Josh, you're an electrical engineer by training, but your career has taken you from refinery operations to midstream project management and now large-scale power generation. Looking back, what drew you to such a broad range of roles across the energy industry?
0:49 Yeah, I would say uh early on in uh my my college career yeah, I had the opportunity to co-op at an oil refinery uh for marathon. And during that co-op, I fell in love with the complexity, just that uh refineries have in them. And then it's kind of a nice mix. I think in that in general, uh it's a good mix of like white collar and blue collar, kind of getting out the people and seeing different equipment rather than behind a desk. I would say just say just like oil and gas and being a power generator generator. Then just opportunities and all those things.
1:28 Yep, absolutely. So today you're helping develop a power generation projects ranging from a few megawatts to an entire gigawatt, much of it supporting hyperscale data centers. How's that explosion of AI and cloud computing changed the work you're doing today?
1:43 I'd say it's changed quite a bit. 80% of my uh specifically that whether that was refining or uh upstream and midstream operations and capital projects, and and just in the last uh 12 to 18 months, the the opportunity to stop maybe carrying parts about uh the molecules and what's happening there and there and electronics has just been just been kind of nuts with the different types of project opportunities that are coming in. Um so it's definitely a bad impact. The work that I'm doing. I would say electrical engineering is a lot more prevalent for sure, you know, on all this power generation side of versus maybe what uh the electrical engineer contributes to uh oil and gas. So it's definitely yeah, absolutely.
2:36 So you've made a point that I think most people never consider. We spend all day on our phones and online, but very few people stop to think about the energy required to keep that digital ward digital world running. What do you wish people understood about that connection?
2:52 Yeah, so I think things just as you mentioned, you mentioned people don't realize I guess people are familiar with like the concept. Everyone carries like a carpet carbon claim and you operate your car or you fly on a plane, right? It's a whole hydrocarbon that everyone's kind of familiar with. I think the main people don't also realize that you know using the internet for cloud storage or you know, the phone writing social media apps or really everything we interact with on a daily basis is is now written. I mean, by natural gas, especially in the in the data center lab, right? Uh, I think everybody knows that natural gas is a big uh energy source for those facilities, but I think sometimes they don't think about what what is coming out of facilities. And it doesn't have it to be uh uh machine learning AI or even server hard, data farm or just uh static storage that are storing your your pictures on your phone anything, right? It requires power mob reads. So I think that it has some kind of hard work outside of what typical driving vehicles or that sort of thing, the last five, ten years, ten years of thought.
4:09 Right, right. Yep, absolutely. You bring up a good point, and I think too, that's one thing that we all forget about, and it's easy to forget about, especially, you know, when everybody just takes for granted just being able to go flip a light switch on and not to think about where that power or that energy comes from. And, you know, especially here in the U.S. with the boom of the data centers starting to take off, it's gonna definitely help uh the industry, you know, be able to grow, of course, because it's gonna be a reliable, clean source of power for all these data centers. So one project you highlighted was developing what essentially a modular repeatable power plant design and a sort of build-your-own generation package. What problem was that trying to was that trying to solve?
4:52 Yeah, so everything and maybe data center space data or speed demand group and often speed power. And so you're various uh urban technology reciprocated engines that uh can only get so large for power generated, usually four megawatts or so. And then you add them your natural gas urban engines that those can get uh pretty large. And then you have your combined cycle cycle technology or the natural gas tribute with uh uh heat recovery and steam power generation, which is what your typical utility plants are using. And this particular was was uh for reciprocating engine engines, uh mostly because of the uh availability of reciprocating engine engine uh power generation. And and this project because from a high-level level system, it's not that complicated to uh uh hook up a bunch of different machines to an electrical system like that at and have them operate uh to get to get you a large scale power on the scale of 100 megawatts to two, three, four hundred watts, and you set it up at another megawatts and you could kind of do a limit um as equipment becomes available. Uh we were solving a problem uh specifically for that power hardware bit.
6:24 Yeah, absolutely. So you've worked both for a massive company like Marathon and much smaller engineering firm. How's that experience changed the way you think about decision making and project edge edu well, let me refer the execution, project execution?
6:39 So yeah, definitely. Uh uh working for a company like Marathon, Marathon, very large refining facility, there's a lot of procedures, a lot of sign-off, a lot of checking, because you you can't really have anything over on a refinery, both from a safety standpoint uh and a financial uh at safety is paramount for facilities like that. The whole overall project cost is a lot slower than a lot of planning. And they certainly have a good sort of to throw a lot of models at it to get the job done on whatever timeline required versus being especially upstream and fixed at a smaller consulting. Um the speed of project is just insanely on a different magnitude um from what I would experience at a larger company in a refinery setting. Um decision can go a lot quicker, there's less sign-off and because the naming of the facility that has a little bit less dangerous or refinery, and so not having that bureaucracy of all these specs and approvals help allows you to get you to your constructed facility a lot faster.
7:57 Yeah, absolutely. I completely agree. So one challenge you brought up was attracting the next generation into field operations and skilled technical roles. Why do you think that's becoming such a difficult problem for the industry?
8:09 Yeah, I don't uh at least from my perspective. Uh so I'm I just chart 30 30 that we kind of got all those generations uh social media media people up with it and then um I think you know people get a lot of FOMO might be the term, but um you know, you could see the highlight reel of it in LinkedIn or LinkedIn or whatever and and people see all the work from home jobs uh high high salaries. Uh you know, someone who might be the might be the uh or uh you know job you can you can you can easily become in that have to be committed to understanding you know maybe not compared to other people's situations I think that's an extremely prevalent way it's just it's with social media. So I mean I'm a driver, but I could at least tell these how that can contribute to people trying to drive their decisions and their career, right? It's just what they see on the internet.
9:24 Yeah, absolutely. I completely agree. So we're gonna go ahead and jump to the last question today. So if you could wave a magic wand and fix anything across the industry tomorrow, you said it would be speeding up decision making. Why do slow decisions end up costing companies so much more money than people realize?
9:43 Yeah, I think again, I use the hyperscale industry that I'm currently in. Faster to end service, it's only quicker. Um and I think people are willing to buy money uh to add resources and remove some of the bureaucracy to achieve that faster constructive timeline. I think especially as engines engineered. People like to over airline and go down every kind of basis. And I would say oftentimes it's not necessary to go down all of those different paths, right? If you can design something that's 80% correct correct, that's that's gonna be good enough enough, you know, you know, probably 98% of the time, and that's usually that's usually good enough to make a decision and move on.
10:29 Yeah, absolutely. I I think it's one of those things that we see lots of times when they say I think they call it analysis paralysis. You know, you hear lots of times different terms where you know decisions get slowed down and it definitely makes it tough on big projects and you know, trying to get executed in a timely manner. You know, sometimes, you know, just sitting on a decision for a week or two can slow down projects by months, and people just don't realize that. So you bring up a great point. Well, Josh, I definitely appreciate your time today. I know you're on a time constraint. So uh one of the things we like to do for all of our guests that come on our show and take time out of their busy schedule is send out a thank you package. So we send out a care package that says, hey, thank you for coming on the podcast and being a part of the show. And also, too, we'll get it mailed out. So just when you get a chance, if you can shoot me your home address via email or your office address, and we'll get a care package sent out and we'll let you know when we're gonna get ready to post the show and go live. Uh it'll be uh we'll tag you on LinkedIn and reach out to you, and so that way you can share it to all your friends as well. So we definitely appreciate your time today. And Josh, you have a great rest of your afternoon, buddy.
11:42 Yeah, thanks, David. Thanks, David.
11:44 Thank you. You bet. Bye. Bye-bye.
11:51 This has been another great episode of Under Pressure. Compressor Talk by Midwest Compressor. With your host, David Abshire. Until next time, keep pumping.